Published: “An End-to-End Autonomous UAV System in GPS-Denied and Unstructured Environments” (title incomplete but published).
Maintains a Google Scholar profile listing all publications.
Hosts a “Wall of Certificates” page showcasing achievements.
Background
Research focuses on understanding and interacting with the dynamic 3D world, with a particular emphasis on human behavior.
Combines computer vision, computer graphics, and machine learning to explore how to capture, perceive, and understand the 4D world from 2D images and videos.
Aims to develop systems that learn priors about the 4D world from visual data to build more realistic and interactive human-centered virtual environments.
Currently collaborating with the Baltimore Orioles (MLB) to develop robust monocular 3D human modeling solutions for quantitative baseball player analysis to improve performance, inform strategy, and prevent injuries.
Research interests include dynamic motion segmentation with egomotion compensation, robust puck localization in ice hockey, unique player identification in low-resolution broadcast videos, player action recognition, human pose estimation, and 3D scene recovery from monocular images.